[cairo] Problem with cairo_scaled_font_glyph_extents() on Win32

[Chris Wilson]

On Mon, 2009-02-09 at 15:17 +0100, Elmar Braun wrote:
> Can I expect that the old behaviour will be restored in a future release
> of 1.8? Or should I work around the new behaviour, by checking whether a
> glyph is valid before calling cairo_scaled_font_glyph_extents? (If so:
> any suggestions how the latter can be done with Cairo, without resorting
> to Win32-specific code? Throwing away and immediately re-creating the
> messed up scaled fonts seems kinda wasteful.)

In principle an invalid glyph should be given an invisible (0x0) extent,
and certainly shouldn't be throwing a permanent error unless the
underlying font blew up in the process.

As a general request, even if it seems blindingly obvious, can you
please include a test case with your problem report. It makes diagnosing
the problem that much easier and provides us with a useful regression
test. In the absence of the test case, the best information you can give
us is pinpointing the error by attaching a breakpoint to _cairo_error()
and printing the stack trace.

It looks like we're accumulating a few vital fixes in the 1.8 branch, so
we should start targeting a new stable release. So if you have a
favourite bug (+patch!) you want to see in 1.8, please make some noise!
Thank you.
-ickle